As global supply chains become increasingly complex and customs landscapes change unpredictably, the traditional approach to Harmonized System (HS) product classification is breaking down. According to IDC research, companies using manual classification processes spend 60% more time on compliance activities. Learn how we got here and where we see the industry headed.
Jump to ↓
Era 1: Physical inspection and printed price list (pre-1988)
Era 2: Harmonized Systems Change Everything (1988)
Era 3: The era of digitization and electronic filing (1990s)
Era 4: The advent of rule-based expert systems (late 1990s to early 2000s)
Era 5: The rise of machine learning (2010s)
Period 6: The AI Era — Classification as a Strategic Capability (2018-Present)
Why ONESOURCE Global Trade is built for AI-powered classification
the way to go
Era 1: Physical inspection and printed price list (pre-1988)
For most of modern trade history, classification was a practical and highly localized endeavor. Customs officials and trade experts classified the goods by physically inspecting them and checking them against printed tariff schedules. Tariff schedules are thick country-specific booklets with little resemblance to each other, even across borders. Each country maintained its own nomenclature, and international trade became a patchwork of incompatible norms and interpretations.
Classification relied almost entirely on individual expertise. Errors were frequent, costly, and time-consuming to resolve. Misclassified shipments can result in border delays, unexpected customs assessments, or regulatory penalties, but there are limited recourses and no common standards to appeal to. For trade professionals at the time, classification was less a system than an art form passed down through experience.
Era 2: Harmonized Systems Change Everything (1988)
In 1988, the World Customs Organization introduced the Harmonized Goods Description and Coding System (HS), a single six-digit coding framework that has been adopted by more than 200 countries. This was the first major standardization advance in the history of trade, giving the global trading community a common language for the first time.
However, the process remained manual. Traders and customs brokers worked based on paper-bound HS schedules, manually cross-referencing classification decisions and explanatory notes. Although the shared framework reduced errors caused by mismatched naming conventions, classification accuracy still depended on the knowledge and judgment of individual experts.
Era 3: The era of digitization and electronic filing (1990s)
The 1990s brought about the first meaningful efficiency gains. Customs authorities began digitizing tariff schedules, and early electronic declaration systems, such as the U.S. Customs and Border Protection Automated Commercial System, began to appear. Software tools enabled classifiers to search for codes electronically, and companies began building internal databases of pre-classified items.
Although the process was faster, it was still primarily human-driven. Computers were search tools, not decision-making engines. Trade professionals were still responsible for interpreting ambiguous product descriptions, navigating multi-component products, and correctly applying common rules of interpretation.
Era 4: The advent of rule-based expert systems (late 1990s to early 2000s)
first wave of truth Human-based acceleration Achieved through decision trees and expert system software. These tools encoded general rules of interpretation into structured question flows. The classifier works through a series of branching prompts that guide you to HS code recommendations. Early global trade management (GTM) platforms began to emerge during this period.
Accuracy improved, but the system remained fragile. They performed well with simple, well-described products, but struggled with novel, technically complex products, or ambiguous wording. Product descriptions in languages other than English were particularly problematic. Although these tools reduced the burden on individual classifiers, they could not replace human judgment in the long tail of difficult cases.
Era 5: The rise of machine learning (2010s)
As companies amassed large historical datasets on classified shipments, classification began to be reframed as a text classification problem. That is, given a product description, the HS code is predicted. Early machine learning models (TF-IDF, Naive Bayes) have been replaced by more powerful techniques such as gradient boosting and support vector machines.
Accuracy has improved significantly for common products that are well described in major trading languages. However, the model still had significant blind spots, and long-tail products, technical specifications in languages other than English, and assembly of multiple components continued to pose challenges. The model learned from patterns in past data. They were unable to reason about unfamiliar input like experienced trade experts.
 |
|
Period 6: The AI Era — Classification as a Strategic Capability (2018-Present)
The introduction of transformer-based architectures and large language models (LLMs) was a real turning point. These models can understand nuanced product descriptions, natively process multilingual input, infer multiple product attributes simultaneously, and leverage search augmentation approaches that incorporate HS descriptive notes and previous customs rulings in real time.
Today’s AI classification tools can use confidence scores to suggest code, flag ambiguous cases for human review, and continuously improve fixes over time. Several customs authorities are piloting AI-assisted systems for automated declarations. For trade professionals, this change is not incremental, but transformative.
And the timing couldn’t be more important. According to IDC’s MarketScape 2025: Worldwide Global Trade Management Applications for Manufacturers and Exporters, the global trade environment has become a “dynamic and rapidly changing environment, forcing manufacturers and exporters to develop capabilities to better manage the complexity of this element of their business.” IDC data shows that companies are overpaying duties by about 5% and 20% of shipment delays. may be due to incorrect or incomplete customs preparations. Misclassification is not just a compliance issue, it’s also a profitability issue.
The IDC report also highlights that AI and machine learning capabilities are built into GTM software to help teams “manage the dynamics of this environment more consistently, comprehensively, and accurately.” Product classification has been cited as one of the key use cases for delivering true AI value in global trade.
Why ONESOURCE Global Trade is built for AI-powered classification
This is exactly the problem Thomson Reuters ONESOURCE Global Trade was designed to solve. Named the 2025 IDC MarketScape Leader in Global Trade Management Applications for Manufacturers and Exporters, ONESOURCE Global Trade brings together AI-powered classification, comprehensive trade content, and a connected ecosystem, all on a modern SaaS platform.
The platform’s CoCounsel GenAI Assistant streamlines classification workflows and provides comprehensive insights across trade functions, helping global trade professionals move from reactive compliance to proactive risk management. Our dedicated team of 200 global content specialists supports 220 countries and territories, supports more than 155 million content updates annually, typically within 24 hours, and ensures that AI-assisted classification decisions are based on the latest, trusted data.
That combination is important for global trade professionals managing an increasingly volatile customs landscape. “IDC Research Director Travis Eide said:
“By generating comprehensive, timely and accurate intelligence across the end-to-end supply chain, organizations can assess trade-offs in real-time to improve compliance, identify opportunities to manage and reduce costs, and strengthen supplier relationships.”
ONESOURCE Global Trade is built to operationalize that intelligence, turning classification from a time-consuming manual burden to a strategic competitive advantage.
the way to go
The evolution of trade classification spans 60 years and six distinct eras, each era shaped by available technology and the complexity of the trade environment. Today, AI-powered solutions are closing the gap between the speed of global commerce and the ability of compliance teams to respond.
For trade professionals who still rely on manual processes and traditional rules-based systems, the costs of inaction are increasing. The technology exists today to deliver better outcomes, faster, more accurately, and at scale.
Ready to see what AI-powered trade classification can do for your organization?
Learn more about how Thomson Reuters ONESOURCE Global Classification AI and Trade Investigations AI helps leading manufacturers and exporters turn compliance complexity into competitive advantage.
